In a recent post I discussed about how to define a winning strategic approach to revitalize data center networking: set a long-term big picture, define a data center network reference architecture and model, and create a pragmatic transformational roadmap. Next question is typically about how to execute this strategy and its roadmap steps without disrupting current operations.
In some cases execution is relatively easy. In fact, when a brand new data center is built, legacy constraints are relaxed or even removed. Architectural building blocks can be easily deployed in the new, clean environment, and are not influenced by the physical topology of data rooms. In this case you will only have to resist the temptation of creating a sort of clone of the existing environment for the sake of simplifying migration, as operational and economic disadvantages would largely overcome advantages.
If a new environment is not possible, the challenge is to deploy new building blocks and related services without impact on the existing environment and while taking into consideration live changes. Again, you surely will need a preliminary simplification of the “as is” environment, getting rid of legacy configurations, protocols, schemas, and proprietary implementations preventing migration. Every configuration change needed to clean the existing environment must be carefully planned and tracked. This was a key success factor of the strategy that HP IT adopted to migrate its six data centers from Cisco to HP gear with zero downtime.
In all cases, and especially when the legacy footprint is highly complex, a good approach is to adopt a surrounding strategy. The evolution of application and service domains can be leveraged to progressively deploy related networking domains based on the new strategic architectural model. They can then be implemented on clean building blocks, as their dependence on existing configuration and topology can be carefully reviewed and even removed. Exceptions should be avoided unless strictly necessary, to avoid corrupting design cleanness and planned benefits.
This approach brings two major advantages. First of all, because new service requirements are challenging for the legacy environment, it’s better to move quickly into the new strategic design to fully leverage technological and architectural opportunities. Moreover, the impact on the “as is” environment is limited or even null, despite progressive reduction of its footprint as it gets shifted to more modern architecture. New domains will likely correspond to areas of investment targeted by the business functions, or there would have been no reason for the change. The combined effect will be an accelerated growth path for your next-generation architecture.
I will touch on this specific topic in my presentation TB2090 “Network Reference Architecture” at HP Discover 2012 in Las Vegas. Come and meet me in the Discover Zone or a Meet the Expert session, or book your one hour introductory session for the Network Transformation Experience Workshop.